The Wellcome Trust and the Urban Graveyard Effect

In the first two weeks of August there was a great kerfuffle about a BBC educational cartoon which showed a couple in Roman Britain who would be called multiracial in Late Capitalist Britain. Angry essays were typed, tweets flew with the wrath of the Stymphalian Birds, and many people hurried to let the Internet know which faction they aligned with. Neville Morley did a good job of summarizing how most ancient historians think about the problem in his blog post Diversitas et Multicultaralismus (no, a dark-skinned official and his light-skinned wife would not have been unheard of at Bath or Hadrian’s Wall; genetic data is exciting but just one of many kinds of evidence which historians draw upon to understand the past; genes are only loosely connected to identity). The Romans could be horrible snobs and bigots, but most of their stereotypes and slurs were directed at people from other parts of Europe and the Mediterranean … they do not seem to have been very interested in whether people had dark skin and kinky hair. In this post, I would like to talk about one of the methodological questions I have after reading the Wellcome Trust paper from 2015 by Leslie et al. which some people have been citing as evidence that negligible numbers of people from Africa had children in Britain before the 20th century (doi:10.1038/nature14230).

This paper used people born in the rural UK whose grandparents were all born within 80 km of one another to reconstruct the genes of the population of the UK in the 1880s, and then compared this with samples from other parts of the world to reconstruct the movements of populations. The authors paid a lot of attention to mass migration from overseas recorded in histories, but in the agrarian world, one of the most common kinds of mobility was from country to city. A struggling farmer did not have much incentive to move from his farm to a distant one where he had no property or relatives or familiarity with the soil and the climate, but he did have incentives to move to town (it was easier to find paid work, nobody was interested in enforcing debts or service obligation from his old village, and he could hide from old enemies in the crowd). And in recent times when we have good data but before modern public heath, cities tended to have higher death rates than birth rates. Food was expensive, sanitation was poor, and violence was everywhere. The poor starved, or lost all their children to diseases, or were impaled on stakes by invaders, and they were replaced by a stream of hopeful people from the countryside. So a small farmer in Cornwall in 1001 CE is much more likely to have modern descendants than a journeyman shoemaker in York in the same year.

This is a problem, because immigrants tend to cluster in big cities. Whether they come to hold a government office, or practice a specialized trade, or trade local goods for the products of their homeland, or simply get off the boat and start looking for work, they have better opportunities in cities than in the countryside. As a Canadian I am very aware of this because of our own history, but immigration to Canada since 1967 follows patterns which are much older. (For example, the Royal Navy tended to deposit freed slaves in London, and black sailors from the Royal Navy also tended to hang around ports when they retired … a decade ago in a course on the Atlantic slave trade I learned that there was a noticeable black population in 18th and early 19th century London which vanished because they were mostly men and their dating pool was mostly people of European descent).

I suspect that in any time in the last 2000 years, the proportion of people from overseas within a mile of Tower Hill is much higher than the proportion in a random one-mile radius of Yorkshire farmland, but before clean water, food safety inspectors, and vaccinations the farmers were more likely to leave descendants than the city folks.

If the “deadly city” effect was true in antiquity, then the sample collected by the Wellcome Trust was biased towards people descended from places where not many people from overseas lived. Therefore it probably understates the proportion of the population of the UK which arrived from overseas. And if you are on Haida Gwai or in a Mennonite village in Saskatchewan, statistics about the total genetic makeup of the population of Canada may not tell you much about the genetics of your neighbours …

There are many similar events in history. Groups like the Jews have been expelled from England every so often, only to return. The Wellcome Trust study did not find many traces of the Norse settlers, but the areas where they settled were Harrowed by William I and then again the scene of intense fighting and terrible atrocities in the 14th, 15th, and 16th centuries along the Anglo-Scottish Border. On the continent, the populations of whole provinces were deported or massacred in the horrible years from the 1910s to the 1940s. And the ancient world had its Gastarbeiter and Temporary Foreign Workers, although it did not have bureaucracies which forced them to go home at a set time or prevented them from finding other employment. There are all kinds of reasons why a population which was once prominent can have few biological descendants today.

I don’t have time to look at these genetic studies closely … I have a dissertation to finish, and frankly I am not all that interested in who was having whose babies. But do any of my readers know how these studies which extrapolate back from genes in modern populations deal with this issue? The Wellcome Trust study is framed as an analysis of how the genes of modern Britons vary across the country, not an analysis of prehistory.

Further Reading: Another piece by a geneticist which is focused on how genetic evidence should be interpreted not how right-minded people should talk about the past is Jennifer Raff, “If Africans Were in Roman Britain Why Don’t We See Their DNA Today?”, The Guardian 9 August 2017

A paper by an actual expert in this effect and its history is Migration and Urban Graveyards: Comparing Mortality Risks between Urban In-Migrants and Natives in a Western European Port City: The Case of Antwerp, 1846-1920